An Efficient Method for Modeling the Magnetic Field Emissions of Power Electronic Equipment From Magnetic Near Field Measurements

Abstract

International audienc

Electromagnetic Time Reversal in the Near Field: Characterization of Transient Disturbances in Power Electronics

“…16, we observe the presence of several harmonics over a frequency range that covers tens of megahertz. In order to acquire an equivalent model using the frequency inverse method, we have employed the standard GA as an optimization method [8][9], which is a widely adopted identification method in the literature. Therefore, the GA-based method has to be performed for several frequency components of the spectrum, which is a time consuming iterative process for both measurement and computations.…”

“…For instance, unlike the far field technique, it can be performed to estimate the current distribution on microstrip traces in a printed circuit board in the high frequency or for faults locations [3][4]. In order to evaluate EM radiation in the near field region, modeling methods based on inverse problem resolution have been widely presented in the literature [3], [5][6][7][8][9][10][11]. Indeed, EMC NF measurements for inverse problems are performed in the reactive region and are generally interested in the identification of real or equivalent radiating sources in the studied structure [3], [5][6][7][8][9][10][11][12][13][14].…”

“…In this regard, the EM inverse problem has been first established in the frequency domain (FD) to estimate radiated disturbances, using a set of equivalent dipoles [2], [5][6][7][8][9]. For instance, several works have proposed a matrix inversion method to obtain equivalent parameters, which needs a linear system resolution [5][6][7].…”

“…Despite its efficiency, this method needs a large set of measurements and leads to several conditioning problems. Thus, optimization methods, namely genetic algorithms (GA), have been introduced [8][9] to reduce the number of dipoles and obtain a 3D model of the device under test (DUT), but this process needs a heavy computational time. Therefore, authors in [9] have investigated the use of hybrid methods, based on GA, pseudo zernike moment invariant (PZMI), artificial neural networks (ANN), in order to avoid previous limitations and speed up the convergence of the algorithm at a fixed operating frequency.…”

A full time domain methodology based on near field time reversal for equivalent source identification

“…This methodology was improved, in [4], by use of GA in the FD in order to accelerate the convergence of the algorithm. To enhance the computational time, Benyoubi et al [5], proposed a fast technique based on the matrix inversion and an optimization procedure, which allows finding equivalent sources with a smaller number of dipoles, compared to [6]. In [7], FD EM inverse problem is elaborated to evaluate the shielding effectiveness of perforated plates.…”